CN110535159A - A kind of method and system of scale energy-accumulating power station running unit fault pre-alarming - Google Patents
A kind of method and system of scale energy-accumulating power station running unit fault pre-alarming Download PDFInfo
- Publication number
- CN110535159A CN110535159A CN201910676678.4A CN201910676678A CN110535159A CN 110535159 A CN110535159 A CN 110535159A CN 201910676678 A CN201910676678 A CN 201910676678A CN 110535159 A CN110535159 A CN 110535159A
- Authority
- CN
- China
- Prior art keywords
- energy
- storage
- data
- energy storage
- subsystem
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
- H02J3/28—Arrangements for balancing of the load in a network by storage of energy
- H02J3/32—Arrangements for balancing of the load in a network by storage of energy using batteries with converting means
Landscapes
- Engineering & Computer Science (AREA)
- Power Engineering (AREA)
- Supply And Distribution Of Alternating Current (AREA)
- Charge And Discharge Circuits For Batteries Or The Like (AREA)
Abstract
The invention discloses a kind of method and system of scale energy-accumulating power station running unit fault pre-alarming, and wherein method includes: to acquire the discharge data of energy-storage battery as with reference to historical data；The running state parameter data for the energy-storage units that acquisition energy storage total system, energy storage subsystem and energy storage subsystem are included；The analysis model of importance is constructed, the degree of correlation of energy storage subsystem and energy storage total system is analyzed, determines the importance factors of each energy storage subsystem；The degree of correlation for analyzing energy-storage units and energy storage subsystem, determines the importance factors of each energy-storage units；Select energy storage subsystem and energy-storage units that importance factors are greater than preset threshold, the running state parameter data of energy storage subsystem and energy-storage units are analyzed with reference to historical data using long prediction neural network in short-term, the prediction result of future time unit is obtained；When the deviation of energy storage subsystem and energy-storage units prediction result and reference historical data is greater than preset threshold value, fault pre-alarming is issued.
Description
Technical field
The present invention relates to power energy storage technical fields, more particularly, to a kind of scale energy-accumulating power station running unit event
Hinder the method and system of early warning.
Background technique
Energy storage technology is the important component of new era energy system, while being also the key of support new energy networking
Technology.The energy storage industry of China is started late, but development speed is very fast.Currently, domestic energy storage technology is positive in Demonstration Application
Explore the energy storage business application under different scenes, technology, scale and technology path, while specification relevant criterion and detection architecture.
Between 2016~2017 years, China's planning accounts for the 34% of global planning and building scale in the nearly 1.6GW of the energy storage scale built, I
State's energy-storage system, which puts into operation, remains rapid growth.By the end of the year 2017, put into operation the accumulative installation scale of energy storage project in China
28.9GW increases by 19% on a year-on-year basis.The accumulative installation scale of electrochemical energy storage is 389.8MW, increases by 45% on a year-on-year basis, proportion is
1.3%, increase by 0.2 percentage point over the previous year.In all kinds of electrochemical energy storage technologies, the accumulative installation accounting of lithium ion battery
Maximum, specific gravity 58%.
Under the background of the more and more extensive access power grid of this electrochemical energy storage system, for the detection of energy-storage system
And fault pre-alarming technology also becomes the key technology of its Parallel Development.The generating efficiency of energy-storage system is with its capacity attenuation mistake
Journey is impacted serious, while under different discharging efficiencies over time, discharge capability is also different.So
During energy-storage system operation, carry out Time-Series analysis prediction and make fault pre-alarming to be a crucial technical problem.
Summary of the invention
Technical solution of the present invention provides a kind of method and system of scale energy-accumulating power station running unit fault pre-alarming, with solution
The problem of certainly how early warning being carried out to scale energy-accumulating power station running unit failure.
To solve the above-mentioned problems, the present invention provides a kind of sides of scale energy-accumulating power station running unit fault pre-alarming
Method, which comprises
The discharge data under energy-storage battery normal operating condition in battery capacity attenuation process is acquired, by the discharge data
As reference historical data；
Based on large-scale energy-storage system, energy storage total system, energy storage subsystem and the energy storage subsystem are acquired respectively
The running state parameter data for the energy-storage units for being included；
Based on machine learning algorithm, construct the energy storage total system, the energy storage subsystem and the energy-storage units it
Between importance analysis model；The degree of correlation for analyzing the energy storage subsystem Yu the energy storage total system, it is each by judging
The running state parameter of the energy storage subsystem is in each tree in random forest to the operating status of the energy storage total system
Influence, determine the importance factors of each energy storage subsystem；Analyze the energy-storage units and the energy storage subsystem
Degree of correlation, by judging the running state parameter of each energy-storage units in each tree in random forest to the storage
The influence of the operating status of energy subsystem, determines the importance factors of each energy-storage units；
The importance factors for selecting the energy storage subsystem and the energy-storage units respectively are greater than the described of preset threshold
Energy storage subsystem and the energy-storage units, by the running state parameter data and institute of the energy storage subsystem and the energy-storage units
It states and is analyzed with reference to historical data using long prediction neural network in short-term, obtain the prediction result of future time unit；Work as institute
When stating energy storage subsystem and the energy-storage units prediction result and the deviation with reference to historical data greater than preset threshold value, hair
It is out of order early warning.
Preferably, the discharge data under the acquisition energy-storage battery normal operating condition in battery capacity attenuation process, institute
Stating discharge data includes: real-time discharge power, discharge voltage and state-of-charge；
When the data volume of collected discharge data is more than the rating data amount of operating status discharge data, with practical work
The discharge data that the discharge data of condition replaces operating status specified is as new operating status discharge data；
The preset data amount of discharge data is set, when the data volume of collected discharge data is more than the preset data amount
When, then history discharge data is replaced with new discharge data.
Preferably, described to be based on large-scale energy-storage system, energy storage total system, energy storage subsystem and described are acquired respectively
The running state parameter data for the energy-storage units that energy storage subsystem is included include:
Acquire overall operation power, operation total voltage and the overall state-of-charge of the energy storage total system；Described in acquisition
Operation power, working voltage and the state-of-charge of energy storage subsystem；Acquire operation power, the operation electricity of the energy-storage units
Pressure and state-of-charge；
The data of the similar parameter of the energy storage total system, the energy storage subsystem and the energy-storage units are divided
Class generates energy-storage system operation power data library, energy-storage system working voltage database and energy-storage system and runs state-of-charge number
According to library.
Preferably, described to be based on machine learning algorithm, construct the energy storage total system, the energy storage subsystem and described
The analysis model of importance between energy-storage units, further includes:
Based on machine learning algorithm, using random forests algorithm construct the energy storage total system, the energy storage subsystem with
And between the energy-storage units importance analysis model.
Preferably, the discharge data under the acquisition energy-storage battery normal operating condition in battery capacity attenuation process, will
The discharge data, which is used as, refers to historical data, further includes:
Battery capacity decays according to the friction speed rate of decay.
Based on another aspect of the present invention, a kind of system of scale energy-accumulating power station running unit fault pre-alarming, institute are provided
The system of stating includes:
First acquisition unit, for acquiring the discharge count under energy-storage battery normal operating condition in battery capacity attenuation process
According to using the discharge data as with reference to historical data；
Second acquisition unit, for be based on large-scale energy-storage system, respectively acquire energy storage total system, energy storage subsystem with
And the running state parameter data of the energy storage subsystem energy-storage units that are included；
Construction unit constructs the energy storage total system, the energy storage subsystem and institute for being based on machine learning algorithm
State the analysis model of importance between energy-storage units；The degree of correlation of the energy storage subsystem Yu the energy storage total system is analyzed,
By judging that the running state parameter of each energy storage subsystem is always to the energy storage in each tree in random forest
The influence of the operating status of system determines the importance factors of each energy storage subsystem；Analyze the energy-storage units with it is described
The degree of correlation of energy storage subsystem, by judging the running state parameter of each energy-storage units every in random forest
To the influence of the operating status of the energy storage subsystem on tree, the importance factors of each energy-storage units are determined；
Prewarning unit, the importance factors for selecting the energy storage subsystem and the energy-storage units respectively are greater than pre-
If the energy storage subsystem of threshold value and the energy-storage units, by the operating status of the energy storage subsystem and the energy-storage units
Supplemental characteristic is analyzed with reference to historical data using long prediction neural network in short-term with described, obtains the pre- of future time unit
Survey result；It is preset when the energy storage subsystem and the energy-storage units prediction result and the deviation with reference to historical data are greater than
Threshold value when, issue fault pre-alarming.
Preferably, first acquisition unit is for acquiring battery capacity attenuation process under energy-storage battery normal operating condition
In discharge data, the discharge data includes: real-time discharge power, discharge voltage and state-of-charge；
When the data volume of collected discharge data is more than the rating data amount of operating status discharge data, with practical work
The discharge data that the discharge data of condition replaces operating status specified is as new operating status discharge data；
The preset data amount of discharge data is set, when the data volume of collected discharge data is more than the preset data amount
When, then history discharge data is replaced with new discharge data.
Preferably, second acquisition unit is used to be based on large-scale energy-storage system, acquires energy storage total system, storage respectively
Can the running state parameter data of subsystem and the energy storage subsystem energy-storage units that are included include:
Acquire overall operation power, operation total voltage and the overall state-of-charge of the energy storage total system；Described in acquisition
Operation power, working voltage and the state-of-charge of energy storage subsystem；Acquire operation power, the operation electricity of the energy-storage units
Pressure and state-of-charge；
The data of the similar parameter of the energy storage total system, the energy storage subsystem and the energy-storage units are divided
Class generates energy-storage system operation power data library, energy-storage system working voltage database and energy-storage system and runs state-of-charge number
According to library.
Preferably, the construction unit is used to be based on machine learning algorithm, constructs the energy storage total system, energy storage
The analysis model of importance between system and the energy-storage units, further includes:
Based on machine learning algorithm, using random forests algorithm construct the energy storage total system, the energy storage subsystem with
And between the energy-storage units importance analysis model.
Preferably, second acquisition unit is for acquiring battery capacity attenuation process under energy-storage battery normal operating condition
In discharge data, using the discharge data as refer to historical data, further includes:
Battery capacity decays according to the friction speed rate of decay.
Technical solution of the present invention provides a kind of method and system of scale energy-accumulating power station running unit fault pre-alarming, wherein
Method includes: the discharge data acquired under energy-storage battery normal operating condition in battery capacity attenuation process, and discharge data is made
For with reference to historical data；Based on large-scale energy-storage system, energy storage total system, energy storage subsystem and energy storage subsystem are acquired respectively
The running state parameter data of the included energy-storage units of system；Based on machine learning algorithm, energy storage total system, energy storage subsystem are constructed
The analysis model of importance between system and energy-storage units；The degree of correlation for analyzing energy storage subsystem and energy storage total system, passes through
Judge the running state parameter of each energy storage subsystem in each tree in random forest to the operating status of energy storage total system
Influence, determine the importance factors of each energy storage subsystem；The degree of correlation for analyzing energy-storage units and energy storage subsystem, passes through
Judge the running state parameter of each energy-storage units in each tree in random forest to the operating status of energy storage subsystem
It influences, determines the importance factors of each energy-storage units；The importance factors of energy storage subsystem and energy-storage units are selected respectively
Greater than the energy storage subsystem and energy-storage units of preset threshold, by the running state parameter data of energy storage subsystem and energy-storage units with
It is analyzed with reference to historical data using long prediction neural network in short-term, obtains the prediction result of future time unit；Work as energy storage
When the deviation of subsystem and energy-storage units prediction result and reference historical data is greater than preset threshold value, fault pre-alarming is issued.This
Inventive technique scheme considers energy-storage system by the short -board effect shadow of energy-storage units for the accident analysis in large-scale energy storage system
It rings obviously, will lead to whole energy-storage system operating parameter when energy-storage units a certain in energy-storage system break down and larger change occurs
Change.When carrying out fault detection, malfunction elimination is not carried out for physical factor, is main ginseng with energy-storage system running state parameter
It examines, the unit that direct Judging fault occurs in the case where data visualization carries out fault pre-alarming work.
Detailed description of the invention
By reference to the following drawings, exemplary embodiments of the present invention can be more fully understood by:
Fig. 1 is the method flow according to the scale energy-accumulating power station running unit fault pre-alarming of the preferred embodiment for the present invention
Figure；
Fig. 2 is the method flow according to the scale energy-accumulating power station running unit fault pre-alarming of the preferred embodiment for the present invention
Figure；And
Fig. 3 is the system structure according to the scale energy-accumulating power station running unit fault pre-alarming of the preferred embodiment for the present invention
Figure.
Specific embodiment
Exemplary embodiments of the present invention are introduced referring now to the drawings, however, the present invention can use many different shapes
Formula is implemented, and is not limited to the embodiment described herein, and to provide these embodiments be at large and fully disclose
The present invention, and the scope of the present invention is sufficiently conveyed to person of ordinary skill in the field.Show for what is be illustrated in the accompanying drawings
Term in example property embodiment is not limitation of the invention.In the accompanying drawings, identical cells/elements use identical attached
Icon note.
Unless otherwise indicated, term (including scientific and technical terminology) used herein has person of ordinary skill in the field
It is common to understand meaning.Further it will be understood that with the term that usually used dictionary limits, should be understood as and its
The context of related fields has consistent meaning, and is not construed as Utopian or too formal meaning.
Fig. 1 is the method flow according to the scale energy-accumulating power station running unit fault pre-alarming of the preferred embodiment for the present invention
Figure.The application embodiment is passed through in the process of running based on the laboratory data of energy-storage battery and field working conditions data
Constantly data update the different demands adapted under different situations, avoid presetting related data and standard bring mistake is estimated
Meter sets fault pre-alarming standard according to actual condition demand.It is different for the data acquisition setting of different time scales simultaneously
Data input size.The application is during time series forecasting, using long neural network algorithm in short-term when the biggish fluctuation of generation
When will lead to overall operation process when can only guarantee about tracking trend stabilization, and can not the complete timing tracking in certain error
System.The error that the method balance shake that the application is weighted when larger shake integrally occurs generates, in the mistake of certain error
Under poor standard, when macro-forecast trend remain to be consistent with actual condition state then with respective weights reduce actual value and predicted value it
Between gap.As shown in Figure 1, a kind of method of scale energy-accumulating power station running unit fault pre-alarming, method include:
Preferably, in step 101: the discharge count under acquisition energy-storage battery normal operating condition in battery capacity attenuation process
According to using discharge data as with reference to historical data.Preferably, battery capacity under energy-storage battery normal operating condition is acquired to decay
Discharge data in journey, discharge data include: real-time discharge power, discharge voltage and state-of-charge；When collected electric discharge
When the data volume of data is more than the rating data amount of operating status discharge data, operation shape is replaced with the discharge data of actual condition
The specified discharge data of state is as new operating status discharge data；The preset data amount of discharge data is set, when collected
When the data volume of discharge data is more than preset data amount, then history discharge data is replaced with new discharge data.Preferably, it acquires
Discharge data under energy-storage battery normal operating condition in battery capacity attenuation process, using discharge data as with reference to history number
According to, further includes: battery capacity decays according to the friction speed rate of decay.
The laboratory of the application battery according to used in energy-storage system and factory state operating parameter are as historical data, with reality
(such as discharge rate can be with are as follows: 1C, 1.5C, 2.5C, the application can be by reality for a variety of different discharge rates under the duty requirements of border
Demand is selected) carry out energy-storage battery normal operating condition under with capacity attenuation running state parameter rating data collection, packet
Include the real-time discharge power (P in discharge processunit), discharge voltage (Vunit) and state-of-charge (SOCunit)；
The application is integrated using running state parameter rating data as standard, using the polynomial regression algorithm in machine learning,
Energy-storage battery is drawn with the operating status trend curve of capacity attenuation, and exports parameter and intercept parameter in its corresponding weight
(w1,...wn,b)；
The application is during actual condition is run, when the data volume that collected system operates normally data set is more than
When the data volume of operating status rating data collection, replace operating status rating data collection as new fortune using actual condition data set
Row state rating data collection, concurrently sets a rated capacity, whenever new data capacity is more than this setting capacity, then with new
Data set replace history data set.
Preferably, in step 102: be based on large-scale energy-storage system, respectively acquire energy storage total system, energy storage subsystem with
And the running state parameter data of the energy storage subsystem energy-storage units that are included.It is preferably based on large-scale energy-storage system, point
Not Cai Ji energy storage total system, energy storage subsystem and energy storage the subsystem energy-storage units that are included running state parameter data packet
It includes: overall operation power, operation total voltage and the overall state-of-charge of acquisition energy storage total system；Acquire energy storage subsystem
Run power, working voltage and state-of-charge；Acquire operation power, working voltage and the state-of-charge of energy-storage units；
The data of the similar parameter of energy storage total system, energy storage subsystem and energy-storage units are classified, energy-storage system is generated and runs function
Rate database, energy-storage system working voltage database and energy-storage system run state-of-charge database.
The application is directed to scale electrochemical energy storage power station, is respectively formed energy storage total system (PCS), energy storage subsystem
(PCSn) and the running state parameter database of the energy storage subsystem energy-storage units (BMSnn) that are included.The application is directed to scale
Change the process that electrochemical energy storage power station forms running state parameter database are as follows:
The application is directed to the running state parameter in scale electrochemical energy storage power station, forms multi-stage data acquisition storage system
System, is respectively formed energy storage total system (PCS), the energy-storage units that energy storage subsystem (PCSn) and energy storage subsystem are included
(BMSnn) running state parameter database.
The application acquires in overall large-scale energy storage system operational process respectively during acquiring energy-storage system data
State parameter, such as overall operation power (Ptotal), operation total voltage (Vtotal), overall state-of-charge (SOCtotal) etc., with
And the operation power (P of each monomer energy-storage system of operationn), working voltage (Vn), state-of-charge (SOCn) and subordinate it is each
Operating status power (Pnn), the working voltage (V of unitnn), state-of-charge (SOCn), and by corresponding energy storage total system, energy storage
Subsystem, the same parameter data classification and then formation, energy-storage system of corresponding energy-storage units run power data library (DataP)、
Energy-storage system working voltage database (DataV), energy-storage system run state-of-charge database (Datasoc)。
Preferably, in step 103: being based on machine learning algorithm, building energy storage total system, energy storage subsystem and energy storage list
The analysis model of importance between member；The degree of correlation for analyzing energy storage subsystem and energy storage total system, by judging each energy storage
Influence of the running state parameter of subsystem in each tree in random forest to the operating status of energy storage total system determines every
The importance factors of a energy storage subsystem；The degree of correlation for analyzing energy-storage units and energy storage subsystem, by judging each energy storage
Influence of the running state parameter of unit in each tree in random forest to the operating status of energy storage subsystem determines each
The importance factors of energy-storage units.
Preferably, in step 104: the importance factors for selecting energy storage subsystem and energy-storage units respectively are greater than default threshold
The energy storage subsystem and energy-storage units of value by the running state parameter data of energy storage subsystem and energy-storage units and refer to history number
It is analyzed according to using long prediction neural network in short-term, obtains the prediction result of future time unit；When energy storage subsystem and storage
When the deviation of energy unit prediction result and reference historical data is greater than preset threshold value, fault pre-alarming is issued.
It is preferably based on machine learning algorithm, it is important between building energy storage total system, energy storage subsystem and energy-storage units
Property analysis model, further includes: be based on machine learning algorithm, using random forests algorithm construct energy storage total system, energy storage subsystem
The analysis model of importance between system and energy-storage units.
The application according to influence of each unit to overall operation parameter in importance analysis function real-time analyzer, according to
The process of influence of each unit to overall operation parameter in importance analysis function real-time analyzer are as follows:
The application acquires in overall large-scale energy storage system operational process respectively during acquiring energy-storage system data
State parameter, such as overall operation power (Ptotal), operation total voltage (Vtotal), overall state-of-charge (SOCtotal) etc., with
And the operation power (P of each monomer energy-storage system of operationn), working voltage (Vn), state-of-charge (SOCn) and subordinate it is each
Operating status power (Pnn), the working voltage (V of unitnn), state-of-charge (SOCn), and by corresponding energy storage total system, energy storage
Subsystem, the same parameter data classification and then formation, energy-storage system of corresponding energy-storage units run power data library (DataP)、
Energy-storage system working voltage database (DataV), energy-storage system run state-of-charge database (Datasoc)。
The application constructs the fortune that each system operational parameters are directed to total system according to the random forests algorithm in machine learning
The importance function of row parameter, while each energy-storage units are constructed to the operating parameter importance analysis function of its said system.Shape
At importance histogram, the influence of each unit and each system operational parameters to total system is analyzed, is not less than with weighing factor
80% is reference, chooses corresponding great influence unit.
The application is according to importance analysis as a result, carrying out timing using operating status of the time sequence forecasting method to energy-storage units
Forecast analysis, and warning information is provided.As shown in Fig. 2, the application is using time sequence forecasting method to the operating status of energy-storage units
Time series forecasting analysis is carried out, and the process for providing warning information is as follows:
The running state parameter timing curve for establishing energy storage total system establishes the running state parameter timing of energy storage subsystem
Curve.Acquired results are analyzed according to importance function simultaneously, the state parameter timing for establishing great influence unit and system compares
Analysis chart carries out intuitive comparative evaluation to the analysis result that importance influences；
According to the historical data of importance analysis result, system and unit energy storage, long prediction neural network pair in short-term is selected
Data carry out real-time forecast analysis, and the application is directed to second level sampled data, using 1 to 2min data as input data, under prediction
The output data of one chronomere；
Prediction curve and true operating condition are subjected to real time contrast's analysis, when will lead to when larger fluctuation occurs in system
There is relatively large deviation for situation in sequence prediction.According to corresponding deviation amplitude, actual prediction demand can achieve in general trend
In the case where, corresponding weighting processing is carried out to system prediction result, when being unable to satisfy overall trend, corrective networks structure is simultaneously
Update the training dataset of input.It concurrently sets and Alert Standard is set according to actual condition, it is complete within prediction result 1-2min
Portion then carries out fault pre-alarming lower than standard.
The application embodiment provides a kind of scale distributed electrical chemical energy storage power failure method for early warning, is based on storage
Based on a variety of data such as energy battery factory nominal parameter and the running state parameter of field working conditions, mentioned using intelligent algorithm
The importance parameter for evidence of fetching, and the energy-storage units according to tracking and early warning needed for this selection, consider in energy-storage system power generation process
A certain energy-storage units can the influence caused by whole energy-storage system when breaking down.It is at any time according to energy-storing and power-generating system simultaneously
The characteristic of existing performance degradation, using Time Series Analysis Method to the running state parameter progress time series forecasting of energy-storage system, and according to
Practical operation situation and data acquisition time scale determine the early warning value within the scope of certain time.Pass through this fault pre-alarming mode
The unit that may be broken down and corresponding malfunction parameter can be prejudged in advance in the process of running, protected by early warning mechanism
Demonstrate,prove the operation of energy-storage system safety and steady.
Fig. 3 is the system structure according to the scale energy-accumulating power station running unit fault pre-alarming of the preferred embodiment for the present invention
Figure.As shown in figure 3, a kind of system of scale energy-accumulating power station running unit fault pre-alarming, system include:
First acquisition unit 301, for acquiring putting in battery capacity attenuation process under energy-storage battery normal operating condition
Electric data, using discharge data as with reference to historical data.Preferably, the first acquisition unit is for acquiring energy-storage battery normal operation
Discharge data under state in battery capacity attenuation process, discharge data include: real-time discharge power, discharge voltage, Yi Jihe
Electricity condition；When the data volume of collected discharge data is more than the rating data amount of operating status discharge data, with practical work
The discharge data that the discharge data of condition replaces operating status specified is as new operating status discharge data；Discharge data is set
Preset data amount then replaces going through when the data volume of collected discharge data is more than preset data amount with new discharge data
History discharge data.
Second acquisition unit 302 acquires energy storage total system, energy storage subsystem for being based on large-scale energy-storage system respectively
The running state parameter data for the energy-storage units that system and energy storage subsystem are included.Preferably, the second acquisition unit 302 is used for
Based on large-scale energy-storage system, energy storage total system is acquired respectively, the energy storage that energy storage subsystem and energy storage subsystem are included
The running state parameter data of unit include: the overall operation power for acquiring energy storage total system, operation total voltage and overall lotus
Electricity condition；Acquire operation power, working voltage and the state-of-charge of energy storage subsystem；The operation power of acquisition energy-storage units,
Working voltage and state-of-charge；The data of the similar parameter of energy storage total system, energy storage subsystem and energy-storage units are divided
Class generates energy-storage system operation power data library, energy-storage system working voltage database and energy-storage system and runs state-of-charge number
According to library.
Preferably, the second acquisition unit 302 is for acquiring battery capacity attenuation process under energy-storage battery normal operating condition
In discharge data, using discharge data as refer to historical data, further includes: battery capacity according to the friction speed rate of decay into
Row decaying.
Construction unit 303, for being based on machine learning algorithm, building energy storage total system, energy storage subsystem and energy storage list
The analysis model of importance between member；The degree of correlation for analyzing energy storage subsystem and energy storage total system, by judging each energy storage
Influence of the running state parameter of subsystem in each tree in random forest to the operating status of energy storage total system determines every
The importance factors of a energy storage subsystem；The degree of correlation for analyzing energy-storage units and energy storage subsystem, by judging each energy storage
Influence of the running state parameter of unit in each tree in random forest to the operating status of energy storage subsystem determines each
The importance factors of energy-storage units.Preferably, construction unit 303 is used to be based on machine learning algorithm, building energy storage total system, storage
The analysis model of importance between energy subsystem and energy-storage units, further includes: machine learning algorithm is based on, using random forest
The analysis model of importance between algorithm building energy storage total system, energy storage subsystem and energy-storage units.
Prewarning unit 304, the importance factors for selecting energy storage subsystem and energy-storage units respectively are greater than default threshold
The energy storage subsystem and energy-storage units of value by the running state parameter data of energy storage subsystem and energy-storage units and refer to history number
It is analyzed according to using long prediction neural network in short-term, obtains the prediction result of future time unit；When energy storage subsystem and storage
When the deviation of energy unit prediction result and reference historical data is greater than preset threshold value, fault pre-alarming is issued.
The system 300 and the present invention of the scale energy-accumulating power station running unit fault pre-alarming of the preferred embodiment for the present invention are excellent
It selects the method 100 of the scale energy-accumulating power station running unit fault pre-alarming of embodiment corresponding, is no longer repeated herein.
The present invention is described by reference to a small amount of embodiment.However, it is known in those skilled in the art, as
Defined by subsidiary Patent right requirement, in addition to the present invention other embodiments disclosed above equally fall in it is of the invention
In range.
Normally, all terms used in the claims are all solved according to them in the common meaning of technical field
It releases, unless in addition clearly being defined wherein.All references " one/described/be somebody's turn to do [device, component etc.] " are all opened ground
At least one example being construed in described device, component etc., unless otherwise expressly specified.Any method disclosed herein
Step need not all be run with disclosed accurate sequence, unless explicitly stated otherwise.
Claims (10)
1. a kind of method of scale energy-accumulating power station running unit fault pre-alarming, which comprises
Acquire the discharge data under energy-storage battery normal operating condition in battery capacity attenuation process, using the discharge data as
With reference to historical data；
Based on large-scale energy-storage system, energy storage total system, energy storage subsystem and the energy storage subsystem are acquired respectively and is wrapped
The running state parameter data of the energy-storage units contained；
Based on machine learning algorithm, weight between the energy storage total system, the energy storage subsystem and the energy-storage units is constructed
The analysis model for the property wanted；The degree of correlation for analyzing the energy storage subsystem Yu the energy storage total system, it is each described by judging
The running state parameter of energy storage subsystem is in each tree in random forest to the shadow of the operating status of the energy storage total system
It rings, determines the importance factors of each energy storage subsystem；It is related to the energy storage subsystem to analyze the energy-storage units
Degree, by judging the running state parameter of each energy-storage units in each tree in random forest to energy storage
The influence of the operating status of system determines the importance factors of each energy-storage units；
The energy storage that the energy storage subsystem is greater than preset threshold with the importance factors of the energy-storage units is selected respectively
Subsystem and the energy-storage units, by the running state parameter data of the energy storage subsystem and the energy-storage units and the ginseng
It examines historical data to be analyzed using long prediction neural network in short-term, obtains the prediction result of future time unit；When the storage
When energy subsystem and the energy-storage units prediction result and the deviation with reference to historical data are greater than preset threshold value, event is issued
Hinder early warning.
2. according to the method described in claim 1, battery capacity attenuation process under the acquisition energy-storage battery normal operating condition
In discharge data, the discharge data includes: real-time discharge power, discharge voltage and state-of-charge；
When the data volume of collected discharge data is more than the rating data amount of operating status discharge data, with actual condition
The discharge data that discharge data replaces operating status specified is as new operating status discharge data；
The preset data amount of discharge data is set, when the data volume of collected discharge data is more than the preset data amount,
Then history discharge data is replaced with new discharge data.
3. acquiring energy storage total system, storage respectively according to the method described in claim 1, described be based on large-scale energy-storage system
Can the running state parameter data of subsystem and the energy storage subsystem energy-storage units that are included include:
Acquire overall operation power, operation total voltage and the overall state-of-charge of the energy storage total system；Acquire the energy storage
Operation power, working voltage and the state-of-charge of subsystem；Acquire the operation power of the energy-storage units, working voltage, with
And state-of-charge；
The data of the similar parameter of the energy storage total system, the energy storage subsystem and the energy-storage units are classified, it is raw
State-of-charge database is run at energy-storage system operation power data library, energy-storage system working voltage database and energy-storage system.
4. constructing the energy storage total system, the storage according to the method described in claim 1, described be based on machine learning algorithm
The analysis model of importance between energy subsystem and the energy-storage units, further includes:
Based on machine learning algorithm, the energy storage total system, the energy storage subsystem and institute are constructed using random forests algorithm
State the analysis model of importance between energy-storage units.
5. according to the method described in claim 1, battery capacity attenuation process under the acquisition energy-storage battery normal operating condition
In discharge data, using the discharge data as refer to historical data, further includes:
Battery capacity decays according to the friction speed rate of decay.
6. a kind of system of scale energy-accumulating power station running unit fault pre-alarming, the system comprises:
First acquisition unit, for acquiring the discharge data under energy-storage battery normal operating condition in battery capacity attenuation process,
Using the discharge data as with reference to historical data；
Second acquisition unit acquires energy storage total system, energy storage subsystem and institute for being based on large-scale energy-storage system respectively
State the running state parameter data for the energy-storage units that energy storage subsystem is included；
Construction unit constructs the energy storage total system, the energy storage subsystem and the storage for being based on machine learning algorithm
The analysis model of importance between energy unit；The degree of correlation for analyzing the energy storage subsystem Yu the energy storage total system, passes through
Judge the running state parameter of each energy storage subsystem in each tree in random forest to the energy storage total system
The influence of operating status determines the importance factors of each energy storage subsystem；Analyze the energy-storage units and the energy storage
The degree of correlation of subsystem, by judging the running state parameter of each energy-storage units in each tree in random forest
Influence to the operating status of the energy storage subsystem determines the importance factors of each energy-storage units；
Prewarning unit, the importance factors for selecting the energy storage subsystem and the energy-storage units respectively are greater than default threshold
The energy storage subsystem and the energy-storage units of value, by the running state parameter of the energy storage subsystem and the energy-storage units
Data are analyzed with reference to historical data using long prediction neural network in short-term with described, obtain the prediction knot of future time unit
Fruit；When the energy storage subsystem and the energy-storage units prediction result and the deviation with reference to historical data are greater than preset threshold
When value, fault pre-alarming is issued.
7. system according to claim 6, first acquisition unit is for acquiring under energy-storage battery normal operating condition
Discharge data in battery capacity attenuation process, the discharge data include: real-time discharge power, discharge voltage and charged
State；
When the data volume of collected discharge data is more than the rating data amount of operating status discharge data, with actual condition
The discharge data that discharge data replaces operating status specified is as new operating status discharge data；
The preset data amount of discharge data is set, when the data volume of collected discharge data is more than the preset data amount,
Then history discharge data is replaced with new discharge data.
8. system according to claim 6, second acquisition unit is used to be based on large-scale energy-storage system, adopts respectively
The running state parameter data packet for the energy-storage units that collection energy storage total system, energy storage subsystem and the energy storage subsystem are included
It includes:
Acquire overall operation power, operation total voltage and the overall state-of-charge of the energy storage total system；Acquire the energy storage
Operation power, working voltage and the state-of-charge of subsystem；Acquire the operation power of the energy-storage units, working voltage, with
And state-of-charge；
The data of the similar parameter of the energy storage total system, the energy storage subsystem and the energy-storage units are classified, it is raw
State-of-charge database is run at energy-storage system operation power data library, energy-storage system working voltage database and energy-storage system.
9. system according to claim 6, the construction unit is used to be based on machine learning algorithm, and it is total to construct the energy storage
The analysis model of importance between system, the energy storage subsystem and the energy-storage units, further includes:
Based on machine learning algorithm, the energy storage total system, the energy storage subsystem and institute are constructed using random forests algorithm
State the analysis model of importance between energy-storage units.
10. system according to claim 6, second acquisition unit is for acquiring under energy-storage battery normal operating condition
Discharge data in battery capacity attenuation process, using the discharge data as with reference to historical data, further includes:
Battery capacity decays according to the friction speed rate of decay.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910676678.4A CN110535159B (en) | 2019-07-25 | 2019-07-25 | Method and system for early warning of faults of operation units of large-scale energy storage power station |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910676678.4A CN110535159B (en) | 2019-07-25 | 2019-07-25 | Method and system for early warning of faults of operation units of large-scale energy storage power station |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110535159A true CN110535159A (en) | 2019-12-03 |
| CN110535159B CN110535159B (en) | 2022-04-15 |
Family
ID=68660987
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910676678.4A Active CN110535159B (en) | 2019-07-25 | 2019-07-25 | Method and system for early warning of faults of operation units of large-scale energy storage power station |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110535159B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111126808A (en) * | 2019-12-11 | 2020-05-08 | 天津泰凡科技有限公司 | Data early warning method and device and computer storage medium |
| CN111159224A (en) * | 2020-01-03 | 2020-05-15 | 珠海格力电器股份有限公司 | Energy storage system control method and device and park level energy storage system |
| CN113655391A (en) * | 2021-08-26 | 2021-11-16 | 江苏慧智能源工程技术创新研究院有限公司 | Energy storage power station battery fault diagnosis method based on LightGBM model |
| CN115902646A (en) * | 2023-01-06 | 2023-04-04 | 中国电力科学研究院有限公司 | Energy storage battery fault identification method and system |
| CN115963408A (en) * | 2022-12-19 | 2023-04-14 | 北京双登慧峰聚能科技有限公司 | Fault early warning system and method for single battery of energy storage power station |
| CN116577673A (en) * | 2023-07-12 | 2023-08-11 | 深圳先进储能材料国家工程研究中心有限公司 | Distributed neural network-based energy storage power station fault diagnosis method and system |
| CN117079435A (en) * | 2023-08-25 | 2023-11-17 | 中国能源建设股份有限公司 | Early warning method and system for oil tank of compressed air energy storage power station compressor |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104638772A (en) * | 2013-11-14 | 2015-05-20 | 国家电网公司 | Battery energy storage power station energy management method based on wind power prediction |
| CN105678398A (en) * | 2015-12-24 | 2016-06-15 | 国家电网公司 | Power load forecasting method based on big data technology, and research and application system based on method |
| CN107942255A (en) * | 2017-11-03 | 2018-04-20 | 山东智洋电气股份有限公司 | A kind of storage batteries of transformer substation group state evaluating method based on Data fusion technique |
| CN108445410A (en) * | 2018-04-02 | 2018-08-24 | 国家计算机网络与信息安全管理中心 | A kind of method and device of monitoring accumulator group operating status |
| CN108551167A (en) * | 2018-04-25 | 2018-09-18 | 浙江大学 | A kind of electric power system transient stability method of discrimination based on XGBoost algorithms |
| CN108808671A (en) * | 2018-07-03 | 2018-11-13 | 南京信息工程大学 | A kind of short-term wind speed DATA PROCESSING IN ENSEMBLE PREDICTION SYSTEM method of wind power plant |
| CN108986470A (en) * | 2018-08-20 | 2018-12-11 | 华南理工大学 | The Travel Time Estimation Method of particle swarm algorithm optimization LSTM neural network |
| CN109143085A (en) * | 2018-07-18 | 2019-01-04 | 中国电力科学研究院有限公司 | A kind of method and system carrying out early warning to lithium ion battery based on intelligent algorithm |
| CN109560577A (en) * | 2018-12-04 | 2019-04-02 | 东莞理工学院 | A kind of control method and system of alternating current-direct current mixed distribution formula renewable energy system |
| CN109681391A (en) * | 2017-10-18 | 2019-04-26 | 中车株洲电力机车研究所有限公司 | A kind of blade root bolt fracture fault detection method and medium |
| CN109755949A (en) * | 2019-01-07 | 2019-05-14 | 中国电力科学研究院有限公司 | A kind of optimizing distribution method and device of combined heat and power energy-accumulating power station power |
-
2019
- 2019-07-25 CN CN201910676678.4A patent/CN110535159B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104638772A (en) * | 2013-11-14 | 2015-05-20 | 国家电网公司 | Battery energy storage power station energy management method based on wind power prediction |
| CN105678398A (en) * | 2015-12-24 | 2016-06-15 | 国家电网公司 | Power load forecasting method based on big data technology, and research and application system based on method |
| CN109681391A (en) * | 2017-10-18 | 2019-04-26 | 中车株洲电力机车研究所有限公司 | A kind of blade root bolt fracture fault detection method and medium |
| CN107942255A (en) * | 2017-11-03 | 2018-04-20 | 山东智洋电气股份有限公司 | A kind of storage batteries of transformer substation group state evaluating method based on Data fusion technique |
| CN108445410A (en) * | 2018-04-02 | 2018-08-24 | 国家计算机网络与信息安全管理中心 | A kind of method and device of monitoring accumulator group operating status |
| CN108551167A (en) * | 2018-04-25 | 2018-09-18 | 浙江大学 | A kind of electric power system transient stability method of discrimination based on XGBoost algorithms |
| CN108808671A (en) * | 2018-07-03 | 2018-11-13 | 南京信息工程大学 | A kind of short-term wind speed DATA PROCESSING IN ENSEMBLE PREDICTION SYSTEM method of wind power plant |
| CN109143085A (en) * | 2018-07-18 | 2019-01-04 | 中国电力科学研究院有限公司 | A kind of method and system carrying out early warning to lithium ion battery based on intelligent algorithm |
| CN108986470A (en) * | 2018-08-20 | 2018-12-11 | 华南理工大学 | The Travel Time Estimation Method of particle swarm algorithm optimization LSTM neural network |
| CN109560577A (en) * | 2018-12-04 | 2019-04-02 | 东莞理工学院 | A kind of control method and system of alternating current-direct current mixed distribution formula renewable energy system |
| CN109755949A (en) * | 2019-01-07 | 2019-05-14 | 中国电力科学研究院有限公司 | A kind of optimizing distribution method and device of combined heat and power energy-accumulating power station power |
Non-Patent Citations (5)
| Title |
|---|
| HARISH ANANTHARAMAN: "Modelling an Adaptive e-Learning System Using LSTM and Random Forest Classification", 《2018 IEEE CONFERENCE ON E-LEARNING, E-MANAGEMENT AND E-SERVICES (IC3E)》 * |
| M. KROENER等: "Characterization of energy storage components for a versatile energy management system", 《ENERGY SELF-SUFFICIENT SENSORS 7TH GMM-WORKSHOP》 * |
| 李俊卿等: "基于随机森林重要性的LSTM网络风电功率缺失数据补齐", 《微电网技术》 * |
| 蒋卫祥等: "《大数据时代计算机数据处理技术探究》", 31 March 2019 * |
| 陆继翔等: "基于CNN—LSTM混合神经网络模型的短期负荷预测方法", 《电力系统自动化》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111126808A (en) * | 2019-12-11 | 2020-05-08 | 天津泰凡科技有限公司 | Data early warning method and device and computer storage medium |
| CN111159224A (en) * | 2020-01-03 | 2020-05-15 | 珠海格力电器股份有限公司 | Energy storage system control method and device and park level energy storage system |
| CN113655391A (en) * | 2021-08-26 | 2021-11-16 | 江苏慧智能源工程技术创新研究院有限公司 | Energy storage power station battery fault diagnosis method based on LightGBM model |
| CN115963408A (en) * | 2022-12-19 | 2023-04-14 | 北京双登慧峰聚能科技有限公司 | Fault early warning system and method for single battery of energy storage power station |
| CN115963408B (en) * | 2022-12-19 | 2024-04-16 | 北京双登慧峰聚能科技有限公司 | Energy storage power station single battery fault early warning system and method |
| CN115902646A (en) * | 2023-01-06 | 2023-04-04 | 中国电力科学研究院有限公司 | Energy storage battery fault identification method and system |
| CN115902646B (en) * | 2023-01-06 | 2023-06-13 | 中国电力科学研究院有限公司 | Energy storage battery fault identification method and system |
| CN116577673A (en) * | 2023-07-12 | 2023-08-11 | 深圳先进储能材料国家工程研究中心有限公司 | Distributed neural network-based energy storage power station fault diagnosis method and system |
| CN116577673B (en) * | 2023-07-12 | 2023-09-12 | 深圳先进储能材料国家工程研究中心有限公司 | Distributed neural network-based energy storage power station fault diagnosis method and system |
| CN117079435A (en) * | 2023-08-25 | 2023-11-17 | 中国能源建设股份有限公司 | Early warning method and system for oil tank of compressed air energy storage power station compressor |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110535159B (en) | 2022-04-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110535159A (en) | A kind of method and system of scale energy-accumulating power station running unit fault pre-alarming | |
| CN106154209B (en) | Electrical energy meter fault prediction technique based on decision Tree algorithms | |
| CN108414848A (en) | Electric energy quality multi-period comprehensive early warning method for power distribution network with distributed power supply | |
| CN109409444B (en) | Multivariate power grid fault type discrimination method based on prior probability | |
| CN107153845A (en) | A kind of isolated island detection method of the adaptive grid-connected photovoltaic system based on machine learning | |
| CN109033513A (en) | Method for diagnosing fault of power transformer and diagnosing fault of power transformer device | |
| CN105160459A (en) | Evaluation method for stable operation state of power system | |
| CN102590652B (en) | Electric-information-based equipment performance evaluation system and method | |
| CN108647838B (en) | Power grid situation sensing method based on random matrix theory and online sequence extreme learning machine | |
| CN110348637A (en) | A kind of consideration field-net factor wind-powered electricity generation climbing event method for early warning | |
| CN115760400A (en) | Mining behavior detection method based on electric power data and storage medium | |
| CN113408132B (en) | Reliability evaluation method for regional autonomous power grid considering stochastic fault scene | |
| CN113850017A (en) | System-level fault analysis system and method based on power flow change map | |
| CN106056305A (en) | Power generation system reliability rapid assessment method based on state clustering | |
| CN111160386B (en) | Data-driven power island detection method based on active reactive power disturbance | |
| Lin et al. | A physical-data combined power grid dynamic frequency prediction methodology based on adaptive neuro-fuzzy inference system | |
| CN111965442A (en) | Energy internet fault diagnosis method and device under digital twin environment | |
| Xu et al. | Optimization of emergency load shedding based on cultural particle swarm optimization algorithm | |
| CN110456191A (en) | A kind of method and system of ultra-large battery energy storage power station running unit detection | |
| Hou et al. | Research on real power grid security and stability based on big data analysis method and online data of power grid | |
| CN111598385A (en) | Method and system for determining power utilization behavior based on fuzzy hierarchical analysis and comprehensive evaluation | |
| CN111242175A (en) | Measuring point voltage anomaly detection method and system based on machine learning | |
| Lingang et al. | Research on integrated calculation method of theoretical line loss of MV and LV distribution Network based on Adaboost integrated learning | |
| CN113283702B (en) | Power flow state evaluation method and device integrating safety and stability characteristics of power system | |
| Yang et al. | Synchronous Line Loss Rate Analysis of Transformer District Based on K-Modes Clustering Algorithm and Deep Belief Network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |